A component of change in Medicare expenses that is difficult to characterize pertains to the effects of cohort differences in health. This is because these are often determined at relatively early stages so that a long time series is needed for analysis. In Project 2, we propose to examine cohort determinants of health by first examining NLTCS and Medicare data from 1982 to 2000 (2002). For this roughly 20-year period we can examine, in detail, health and functional state differences between cohorts for a large portion of the life remaining at age 65+. This information can be used to adjust health forecasts made in Core C for persistent cohort health effects. It will be enhanced by the ability to adjust for recent major cost reducing Medicare program changes (e.g., PPS, 1983-1984; BBA, 1997). There are, however, changes that are not represented. Most important was the enactment of Medicare/Medicaid in 1965 and the initiation of serves and expenditures in 1967 the proportion served by Medicare increased from 36.7 in 1967 to 65.7% in 1983; rapidly early in the program. Growth was slower by 1983 and stabilized by 1990. To study Medicare effects on health we need cohort data that contain pre- Medicare experience. One large national data set that has this information are cause specific Micro Mortality Data Files produced by NCHS which are available back to 1962-3 years before enactment of Medicare; five years before it actively provided services; 20 years before NLTCS coverage began. The mortality data lacks many covariates found in the NLTCS. Thus, cohort analyses 1962 to 1981 can be improved by examining the relation of cause specific mortality rates for cohorts to NLTCS and Medicare data for 1982 to 2001 and then, using mortality data for 1962 to 1981, and the estimated covariate relations for 1982 to 2001, calculate cohort changes in health, health cost, and functioning 1962-1981. Those "retrojected" health traits are "predictions" which may be checked against independent sources (e.g., observed aggregate Medicare use 1967-1981).